How Do You Measure the Difference Between Alienation and Estrangement? The Bernet, Gregory, Rohner & Reay (2020) PARQ-Gap Paper

Love Over Exile is a plain-language research and policy archive on parental alienation, written by Malcolm Smith — an alienated parent and author of the forthcoming book Love Over Exile — for non-specialist readers (other alienated parents, family members, therapists, lawyers) who want to understand the evidence base without a psychology qualification or a journal subscription. This page is one entry in that archive.

What the Researchers Asked

When a child resists, refuses, or rejects contact with one parent, evaluators face the same diagnostic question Gardner formulated forty years ago. Is the child rejecting the parent because that parent has been abusive, neglectful, or seriously deficient — what the parental-alienation literature calls realistic estrangement? Or is the child rejecting the parent for no good reason — parental alienation?

The distinction matters because the appropriate response is different in each case. An estranged child needs protection from a harmful parent. An alienated child needs restoration of the relationship the favoured parent has interfered with. Mistake one for the other and the intervention causes harm.

Bernet, Gregory, Rohner and Reay set out in 2020 to test whether a quantitative instrument could draw the line. Their theoretical wager, stated in the abstract, is that “alienated children, who were not abused, tend to engage in splitting and lack ambivalence with respect to their parents; estranged children, who were maltreated, usually perceive their parents in an ambivalent manner.” If that pattern is real, an instrument that measures the gap between a child’s perceptions of each parent should separate the two groups — alienated children showing a large Gap (one parent idealised, the other devalued) and estranged children retaining ambivalence on both.

The hypothesis they tested is narrowly classificatory, not predictive. Verbatim from the abstract: “the hypothesis of this study was that a psychological test — the Parental Acceptance-Rejection Questionnaire (PARQ) — will help to distinguish severely alienated from nonalienated children.”

How the PARQ-Gap Score Was Built — Methodology in Plain English

The paper is a case-control instrument-validation study. It compares two pre-classified groups of children on a derived score from a validated measurement instrument, with the aim of finding a cut-point on the score that separates the groups well.

The instrument is the Parental Acceptance-Rejection Questionnaire (PARQ), developed by co-author Ronald Rohner at the University of Connecticut. The PARQ is a child-report measure of how warm, hostile, indifferent, or rejecting each parent is perceived to be. It is used in 65 languages and over 60 cultures, with 50 years of cross-cultural validation behind Rohner’s broader Interpersonal Acceptance-Rejection Theory. The PARQ was not designed to detect parental alienation.

In this study, each child completed the PARQ twice — once with the mother in mind, and once with the father in mind. From the two scores, the authors derived a new measure: the PARQ-Gap, defined as the difference between a child’s PARQ score for one parent and the PARQ score for the other. Verbatim from the abstract: “the PARQ-Gap score — the difference between each child’s PARQ: Father score and PARQ: Mother score — was introduced and defined in this research.”

The conceptual move is therefore: take a non-PA instrument with strong cross-cultural psychometric backing, derive a new score from it, and use that derived score as a parental-alienation marker. This is both the paper’s elegance and the central locus of methodological debate.

The validation sample comprised 116 children, 45 severely alienated and 71 non-alienated. The 45 alienated children were recruited from the Family Reflections Reunification Program in British Columbia, Canada — a clinic specialising in parental-alienation treatment, founded and clinically directed by co-author Kathleen Reay. The 71 non-alienated children were recruited via Research Match, a US National Institutes of Health volunteer registry.

Crucially, the non-alienated group was mixed: 35 children from intact families, 20 from divorced families maintaining contact with both parents, and 16 children classified as “neglected” via maternal report of paternal unreliability. The article should note this composition carefully — it is the central methodological feature critics return to.

The 99% Accuracy Headline — What It Means and What It Doesn’t

The single most-cited sentence from the paper is the abstract’s headline result:

“Using a PARQ-Gap score of 90 as a cut point, this test was 99% accurate in distinguishing severely alienated from nonalienated children.”

What that means precisely. Among the 116 children in the validation sample, a PARQ-Gap of 90 or more on the difference between the two parents’ scores correctly classified the children into their pre-determined group around 99 times out of 100 — about one misclassified child across the whole sample.

What that result does not mean:

• It is not separately reported sensitivity and specificity. A diagnostic test is normally characterised by two distinct numbers — the true-positive rate among children with the condition (sensitivity) and the true-negative rate among children without (specificity). The PARQ-Gap paper reports a single overall accuracy figure on a sample where the two groups are roughly the same size. Sensitivity, specificity, positive and negative predictive values, and ROC curve analyses are not separately reported.

• It is not a real-world clinical-accuracy figure. Validation-sample accuracy depends on how groups were recruited and on the prevalence balance in the sample. In an unselected clinical population — say, every child referred for a Section 7 welfare report under the Children Act 1989 in the UK, or every custody evaluation in a US state — the prevalence of severe alienation is much lower. The positive predictive value of a 99% accurate test in a low-base-rate population is much lower than 99%.

• The cut-point of 90 is contextual. The body of the paper caveats that a PARQ-Gap of 90 performs best specifically when the favoured parent’s PARQ score is in the 60–70 range — i.e. when the child is idealising the favoured parent strongly. The same Gap of 90 produced by other underlying score configurations may not classify accurately. The 99% figure is conditional on a specific configuration, not the Gap value alone.

What the validation sample actually contained

This is the central feature of the design that the article must report carefully. The comparison group is not a group of children who refused contact for genuine cause. It is a group of children who, broadly, do not refuse contact with either parent — three-quarters of them from families with no separation or with maintained contact post-separation. The contrast is therefore between severe contact refusal and no contact refusal, not between alienation and estrangement in the strict sense.

Read Alongside Mercer (2021) — The Published Critique and Reply

The most important piece of context for this paper is the explicit per-paper critique published the year after — Mercer (2021) in the Journal of Family Trauma, Child Custody & Child Development, titled “Rejecting the idea of rejection as a measure of parental alienation: Comment on Bernet, Gregory, Rohner and Reay (2020).”

Jean Mercer is professor emerita at Stockton University and one of the most consistent published critics of the parental-alienation evidence base. Her 2021 critique of the PARQ-Gap raises six specific objections.

The first is that the PARQ does not measure parental alienation. It measures perceived parental warmth, hostility, indifference, and rejection. Inferring a parental-alienation-specific construct from a derived Gap on a non-PA instrument is theoretically unsupported.

The second is selection bias in the alienated group. The 45 children were recruited from a clinic that defines, treats, and bills for parental alienation. The same conceptual framework that the test is meant to validate determines who counts as a “severely alienated child” in the validation sample.

The third is that the comparison group is not estranged. As detailed above, “non-alienated” mixes intact-family children, divorced-but-contact-maintaining children, and a small “neglected” sub-group — children with no contact-refusal problem at all. The paper’s title promises an alienation-versus-estrangement test; the design tests something narrower.

The fourth is that the Likert-scale construction is psychometrically fragile. The PARQ uses ordinal Likert items; treating their sums as continuous and comparing a difference of two such sums as a numeric “Gap” is psychometrically ambitious. Mercer raises this concern across multiple parental-alienation instruments.

The fifth is that splitting is not diagnostic in adolescents. Some “extreme” emotional polarisation about parents is developmentally normal in adolescence. The paper does not adequately separate splitting-as-pathology from splitting-as-development.

The sixth is that the 99% figure is sample-bound. The accuracy is reported on a sample where group membership was determined by recruitment source, not by an independent gold-standard clinical assessment. It is logically circular to say a clinic-defined group can be classified by a test built around the same defining features.

Bernet, Rohner and Reay (2021) replied in the same journal, titled “Rejecting the rejection of parental alienation: Comment on Mercer (2021).” Their reply addressed Mercer’s claims that the work is pseudoscience, that parental-alienation theory involves false assumptions, that the splitting observed is age-appropriate, and that the conclusions are exaggerated. Both papers are on the published record and the article must acknowledge both — quoting one without the other is not honest engagement.

Comparison table — the PARQ-Gap, the Five-Factor Model, and the Baker Strategies Questionnaire

	PARQ-Gap (Bernet et al. 2020)	Five-Factor Model (Bernet & Greenhill 2022)	Baker Strategies Questionnaire
Where published	J. Forensic Sciences, 65(4)	JAACAP, 61(5)	Multiple papers, 2005–2011
Type	Case-control instrument-validation study	Diagnostic-framework commentary	Behavioural-inventory questionnaire
What it measures	Numeric difference between child’s PARQ scores for each parent	Five clinical criteria for diagnosing PA	Self-report of 17 alienating behaviours by the favoured parent
Headline metric	99% classification accuracy at cut-point 90 (n=116)	All five factors must typically be present for a PA diagnosis	Frequency / strength scoring of each behaviour
Sample	45 alienated + 71 non-alienated children	None — conceptual paper	Adult retrospective samples, varies
Frame strength	Quantitative; uses validated PARQ instrument; cross-cultural backing for the PARQ itself	Specifies what to look for clinically; widely cited	Structured inventory of specific tactics
Frame limitation	Comparison group is not estranged; selection-biased to source clinic; cut-point is contextual	Not empirically validated as a syndromic test; Garber & Simon (2023) argue it is over-specified	Self-report; recall bias; not designed as a diagnostic
What it establishes	A derived score classifies a clinic-recruited alienated group from a mixed non-alienated group with high accuracy	A clinical framework for what parental alienation looks like	A taxonomy of alienating behaviours and their relative frequency
What it does not establish	Real-world diagnostic accuracy; predictive validity over time; PA-vs-estrangement test in the strict sense	An empirically validated diagnostic algorithm	A diagnostic for the child or a measure of harm
Reception	Mercer (2021) explicit critique + Bernet et al. (2021) reply	Garber & Simon (2023) critique; Garber, Simon, Bernet & Griffin (2024) partial walk-back	Used widely in PA practice; less heavily critiqued empirically

Two readings of the table. First, the three instruments work at different layers — the Five-Factor Model on diagnostic structure, the Baker Strategies Questionnaire on behavioural inventory, and the PARQ-Gap on quantitative measurement of the child’s perceptions. Second, none of the three is a standalone diagnostic; they are designed to be used together, and the PARQ-Gap supports — but does not on its own establish — the Five-Factor Model framework.

The article should quote the PARQ-Gap’s 99% headline alongside the explicit Mercer critique in the same paragraph. Either alone, without the other, is misleading.

What This Means for UK Practice

The PARQ-Gap paper is a US/Canadian instrument-validation study. The 45 alienated children were recruited in British Columbia, the 71 non-alienated children in the United States, and the diagnostic-framework discussion is anchored in DSM-5 and the Bernet camp’s terminology. UK readers — particularly UK alienated parents navigating Cafcass and the family courts in England and Wales — need to read the paper through the right adapter.

The UK adapter is the Family Justice Council December 2024 guidance on responding to allegations of alienating behaviour. The FJC explicitly rejects the “parental alienation syndrome” framing, adopts alienating behaviour as the operative term-of-art with a defined three-element test (RRR, AJR, AAA, AB, PB), and is the canonical operating standard for England and Wales family courts.

The relevant UK questions about the PARQ-Gap are not “does it diagnose alienation?” (it does not, even on the authors’ conservative claim) but “is the PARQ-Gap a UK-recognised instrument? Should a UK psychologist deploy it in a Section 7 report?” Three answers:

• The PARQ instrument itself is UK-available. The PARQ exists in 65 languages and is used in UK developmental-psychology research. Any UK psychologist instructed under Family Procedure Rules Part 25 could in principle administer it as part of broader assessment.

• The PARQ-Gap derivation is not UK-endorsed. The Gap derivation is specific to this Bernet team’s 2020 paper and has no separate UK validation, no Cafcass endorsement, and no UK professional-body sign-off. Following Re C [2023] EWHC 345 (Fam) and the FJC December 2024 guidance, UK courts restrict expert witnesses on alienating behaviour to regulated clinical or counselling psychologists with no financial interest in the treatments they recommend. Presenting a 99% accuracy claim from a US/Canadian validation sample as a UK diagnostic standard would not survive proper expert-witness scrutiny.

• The empirical UK floor is in the parallel UK literature. The Hine, Harman, Leder-Elder and Bates (2025) UK prevalence study, in the Journal of Family Violence (DOI 10.1007/s10896-025-00910-4), establishes that 39% of separated UK parents experience alienating behaviours when asked directly and 59% on a 30-item behavioural inventory. UK alienated parents need the prevalence floor, the FJC legal framework, and an honest reading of the international measurement literature — not the 99% headline used uncritically.

Limits, Conflicts of Interest, and the Comparison-Group Caveat

Four honest qualifications belong on every reading of this paper.

The methodological limits the authors partially acknowledge. The body’s limitations section flags the small “neglected” sub-group (n=16) with imprecise classification, the uncertainty about whether the neglected children are actually estranged from their fathers, and the single-clinic recruitment for the alienated group. The cross-cultural validation of the PARQ itself is offered as a strength.

The methodological limits external critics raise. Mercer (2021) and the broader Mercer / Drew / Saini / Faller line of work argue: the PARQ does not measure parental alienation; the alienated group is selection-biased to its source clinic; the comparison group is not estranged; the Likert-scale aggregation across a difference score is psychometrically fragile; splitting is developmentally normal in adolescents; and the 99% figure is sample-bound. Saini, Johnston, Fidler and Bala (2016) used the GRADE quality-assessment tool to evaluate 58 PA empirical studies and concluded that the empirical base for parental alienation is methodologically weak.

The conflict-of-interest disclosures are material and reproduced from the paper. Ronald Rohner is a partner in Rohner Research Publications, which publishes the PARQ instrument the paper validates a derived score from — the senior methodological author has a commercial interest in the underlying instrument. Kathleen Reay founded and clinically directed the Family Reflections Reunification Program — the source clinic for the 45 alienated children. William Bernet receives royalties from Charles C Thomas Publisher (publisher of the Lorandos and Bernet 2020 Parental Alienation — Science and Law textbook) and is founder and president of the Parental Alienation Study Group; Nilgun Gregory received TUBITAK support.

The within-camp partial walk-back. Garber, Simon, Bernet and Griffin (2024) in Family Transitions — a senior-figures consensus-building meeting between Bernet-camp and Garber-Simon-critic figures — concluded that the Five-Factor Model the PARQ-Gap is intended to support “is a good foundation but needs to be more fully developed and defined.” This is a partial walk-back of the strongest version of the strongest claim by senior figures including Bernet himself. It does not retract the PARQ-Gap paper’s classification finding, but it is the most relevant subsequent context for any reader thinking about how to use the instrument.

The honest reading of Bernet, Gregory, Rohner and Reay (2020) is the one this article tries to support. In a case-control validation sample of 116 children — 45 severely alienated from a Canadian PA-treatment clinic, 71 non-alienated from a US volunteer registry — a derived PARQ-Gap score of 90 or more on the difference between the child’s perceptions of each parent classified the two groups with 99% overall accuracy. That is real and it is the strongest available quantitative marker for distinguishing severe alienation from non-alienation in a clinic-recruited validation sample.

It is not a real-world diagnostic test, it is not a UK-recognised forensic instrument, it is not a test of alienation versus genuine estrangement (the comparison group is not estranged), and it is not separately reported as sensitivity and specificity. Read alongside the Mercer (2021) critique, the Bernet (2021) reply, the Saini and colleagues (2016) GRADE assessment, and the Garber, Simon, Bernet and Griffin (2024) walk-back, the picture is layered evidence — a useful starting point for a measurement question that is far from settled, not a finished diagnostic standard.

UK readers — and any reader thinking about how the paper applies to a real custody decision — should also read the FJC December 2024 guidance and the Hine 2025 UK prevalence study. The four papers — the international measurement instrument, the UK legal framework, the UK empirical floor, and the methodology critique — together give a more honest picture than any one of them alone.